---

The joint report by the General Inspectorate of Social Affairs (IGAS), the General Inspectorate of Finance (IGF) and the General Inspectorate of Administration (IGA) analyses the conditions under which artificial intelligence (AI) can bring about lasting improvements to public services, the quality of civil servants’ work and the effectiveness of public policies.

Generative and agent-based AI is emerging as a major driver of transformation in public administration, but its effects will be neither automatic nor uniform.

Its adoption within French public administrations is already underway, but remains recent, varied and often poorly coordinated. The rise of ‘shadow AI’ – the spontaneous and unregulated use of consumer-grade tools by civil servants – reveals a strong appetite for AI, but also exposes public administrations to risks relating to security, reliability, inequalities in access and data protection.

The State and its agencies, local authorities, social security bodies, healthcare institutions and France Travail are progressing at different rates, with uses ranging from general-purpose assistants to specialised sector-specific tools, including support functions.

The report emphasises that the expected benefits are not necessarily quantitative but must be assessed more broadly: reduced processing times, improved service quality, fewer errors, simplified administrative work, better relations with users and an improved quality of life at work.

In the social sector, France Travail, social security organisations and certain healthcare institutions illustrate, through the numerous use cases presented, the potential of AI to facilitate the processing of claims, access to information, the fight against fraud, document management and user support.

Making AI a collective capability for improving public services

However, scaling up requires starting with the actual work carried out by staff. The report emphasises the need for forward-looking workforce and skills planning that incorporates AI, a structured training policy for all staff, managers and specialists, and a genuine social dialogue on technology. This dialogue must enable staff and their representatives to be involved in decisions on how AI is used, in assessing its impact on roles, in risk prevention and in the reallocation of benefits.

The challenge is therefore no longer simply to adopt AI, but to turn it into a collective capacity for improving public services. AI should be viewed less as a technical tool and more as a lever for transforming organisations, skills and processes, serving staff, users and the effectiveness of public policies, and driven by all organisations, not just their digital departments.